4 ms·
Frustratingly none of the coding harnesses seem to default to just firing up a docker container with the working directory mounted and letting the AI run wild i
by londons_explore 3mo ago
Frustratingly none of the coding harnesses seem to default to just firing up a docker container with the working directory mounted and letting the AI run wild inside the container.
- kasey_junk 3mo agoYou can get that working for yourself trivially. My agent environment now has 3 different kinds of compute abstractions, 3 different agents and a nascent workflow engine in it. It works how _i_ want, which is really refreshing.
- londons_explore 3mo agoOh I've built the same too. But it took me a few hours to work out the quirks, when it should just be the default.